Dynamic inventory rationing for systems with multiple demand classes

Abstract

Inventory rationing among demand classes is popular and critical in industries. In the literature most people consider static rationing policies, for the inventory problems with dynamic rationing policies are very complicated or intractable. This dissertation studies dynamic inventory rationing in different circumstances. The research is divided into three parts. The first part considers the systems with Poisson demands and backordering. For the case of zero lead time, we characterize the optimal rationing and ordering policies. For the case of positive lead time, near optimal solution and a lower bound on optimal costs are obtained. Numerical results shows costs of the near optimal solution are close to the optimal costs and our dynamic policy can significantly reduce cost comparing with the static policies in many cases. The second and third parts extend the first part by changing Poisson demand process to general processes, or changing backordering to lost sales.